Modeling progression of screening detected Lung Cancer
We plan to develop a method for simulating, at an individual level, the characteristics of lung cancer in a population at the time of diagnosis, according to the progression and detection of the disease. Lung cancer data from NLST trial will be used to fit the lung cancer progression and detection model. This progression and detection model will be combined with a previously fitted two-stage clonal expansion (TSCE) model of carcinogenesis, previously used in the context of lung cancer and modified and validated by us as part of the CISNET Lung Group’s Smoking Base Case project (Foy et al., 2011).
This TSCE model was fitted using a resampling-based method allowing for estimation of risk factor-dependent parameters from the combination of case-control data and prospective mortality rate data. The progression and detection model will be based in part on another previously developed model which describes individual natural histories of lung cancer and allows estimation of the mortality reduction associated with early detection of lung cancer followed by appropriate treatment.
The model was applied to reconstruct the results of the Mayo Lung Project (MLP) (Flehinger et al. 1993) and was further validated by producing accurate predictions of long-term mortality in the MLP (Gorlova et al, 2001). It also was used to evaluate power in the NLST under reasonable assumptions concerning adherence to the screening regimen and contamination (Kimmel et al, 2005).
We will further modify the model to incorporate genetic susceptibility as approximated by lung cancer family history, available from the NLST, similar to incorporation of mutagen sensitivity as done previously (Gorlova et al, 2003). This way, both the absolute lifetime risk and the distribution of the age at disease onset will become dependent on the genetic susceptibility status. We also will use data on smoking cumulative exposure and on respiratory conditions, such as COPD, to further fine-tune the lifetime susceptibility to lung cancer. Finally, the image description data and image reanalysis as necessary will be used in conjunction with host susceptibility factors to (i) better delineate true lung cancers and (ii) evaluate the association of image characteristics with outcomes among identified lung cancer patients.
Aim 1. To develop a natural history model of lung cancer that will include the dependence of the progression rates (tumor growth and metastasizing) on host factors such as smoking, other exposures, respiratory conditions, and genetic susceptibility (approximated by cancer family history) as well as on initial image characteristics of the tumor.
Aim 2. Based on the model, to identify risk group-specific screening strategies - screening intervals, age of screening initiation, duration of screening (number of lifetime screens) and work-up of positive findings - optimal in terms of potential mortality reduction associated with low-dose CT screening.
Aim 3. To assess the impact of such programs on LC mortality reduction, taking into account screening-associated risks.
Marek Kimmel, PhD, Rice University
Reginald Munden, MD, MD Anderson Cancer Center
Radiologist, MD, TBN